In order to improve the efficiency of road operation and alleviate urban traffic congestion, the urban area of Suzhou city is taken as the research object to carry out the analysis and research of traffic operation state. The data of floating cars are obtained by GPS system, and the data are repaired and preprocessed by using mathematical statistics method. The travel speed and traffic flow are selected as evaluation parameters to construct the calculation model of road travel speed. Agnes clustering algorithm is used to cluster the road flow and average speed, so as to classify the road traffic status and determine the interval value of different grades. The results show that the critical value of serious congestion on main roads in Suzhou is 20km / h, which is lower than the standard value (21km / h); meanwhile, the threshold of moderate and severe congestion of secondary trunk roads is significantly lower than the standard value, which may be due to narrow lanes and mixed traffic of motor vehicles and non motorized vehicles. This study can provide a new method for evaluating urban traffic conditions by using traffic data, improve the understanding of road structure for traffic managers, and have certain reference value for urban road planning anddesign.